Reduced Fcrn-Mediated Transcytosis of Igg2 Due to a Missing Glycine in Its Lower Hinge Received: 19 February 2018 Nigel M

www.nature.com/scientificreports

OPEN Reduced FcRn-mediated transcytosis of IgG2 due to a missing Glycine in its lower hinge Received: 19 February 2018 Nigel M. Stapleton1,2, Maximilian Brinkhaus1, Kathryn L. Armour3,4,6, Arthur E. H. Bentlage1, Accepted: 19 February 2019 Steven W. de Taeye1, A. Robin Temming1, Juk Yee Mok8, Giso Brasser8, Marielle Maas8, Published: xx xx xxxx Wim J. E. van Esch8, Mike R. Clark 3,7, Lorna M. Williamson4,5, C. Ellen van der Schoot1 & Gestur Vidarsson 1

Neonatal Fc-receptor (FcRn), the major histocompatibility complex (MHC) class I-like Fc-receptor, transports immunoglobuline G (IgG) across cell layers, extending IgG half-life in circulation and providing newborns with humoral immunity. IgG1 and IgG2 have similar half-lives, yet IgG2 displays lower foetal than maternal concentration at term, despite all known FcRn binding residues being

preserved between IgG1 and IgG2. We investigated FcRn mediated transcytosis of VH-matched IgG1 and IgG2 and mutated variants thereof lacking Fc-gamma receptor (FcγR) binding in human cells expressing FcRn. We observed that FcγR binding was not required for transport and that FcRn transported less IgG2 than IgG1. Transport of IgG1 with a shortened lower hinge (ΔGly236, absent in germline IgG2), was reduced to levels equivalent to IgG2. Conversely, transport of IgG2 + Gly236 was increased to IgG1 levels. Gly236 is not a contact residue between IgG and FcRn, suggesting that its absence leads to an altered conformation of IgG, possibly due to a less fexible Fab, positioned closer to the Fc portion. This may sterically hinder FcRn binding and transport. We conclude that the lack of Gly236 is sufcient to explain the reduced FcRn-mediated IgG2 transcytosis and accounts for the low maternal/fetal IgG2 ratio at term.

During the frst months afer birth, before infants acquire their own immunity, they are protected by maternal IgG class antibodies, transferred by the neonatal Fc-receptor (FcRn)-mediated transplacental transcytosis1–5. Critically for this transport, FcRn binds IgG with nanomolar afnities at low pH (≤6.5) (found in intracellular vacuoles), while at physiological pH (7.4) this afnity is low6–9. Once it is pinocytosed, the IgG containing vesicle acidifes, causing IgG present to bind to FcRn. Endocytosis and sorting signals in the cytoplasmic tail of FcRn cause FcRn-IgG complexes to be routed away from the lysosomal pathway to the plasma membrane10–16. Upon fusion with the plasma membrane in a series of short events17, the pH returns to physiological levels, IgG-FcRn complexes dissociate, IgG difuses away and FcRn restarts its cycle. Te net result, depending on the route of exocytosis, is either IgG-recycling or transcytosis17–19. In human transplacental transport, IgG is generally thought to be transported from the maternal to fetal circulation in three steps: First IgG is pinocytosed by FcRn expressing syncytiotrophoblasts, it transverses the villus interstitium passively by bulk fow and fnally IgG is transported across fetal villus endothelial cells20. During the frst half of pregnancy, the IgG subclass composition in cord blood resembles that in maternal serum, although the total IgG concentration in cord blood remains lower. At term however the IgG1 concentration in cord blood

1Sanquin Research, Department of Experimental Immunohematology, Amsterdam, The Netherlands, and Landsteiner Laboratory, Amsterdam UMC, University of Amsterdam, Amsterdam, The Netherlands, Plesmanlaan 125, Amsterdam, 1066 CX, The Netherlands. 2Present address: HALIX B.V., J.H. Oortweg 15/17, 2333 CH, Leiden, The Netherlands. 3Department of Pathology, Division of Immunology, University of Cambridge, Tennis Court Road, Cambridge, CB2 1QP, UK. 4Department of Haematology, University of Cambridge, Cambridge, UK. 5NHS Blood and Transplant, Long Road, Cambridge, CB2 2PT, UK. 6Present address: LifeArc, Open Innovation Campus, Stevenage, SG1 2FX, UK. 7Present address: Clark Antibodies Ltd, 10 Wellington Street, Cambridge, CB1 1HW, UK. 8Sanquin Reagents, Amsterdam, Netherlands. Correspondence and requests for materials should be addressed to G.V. (email: [email protected])

Scientific Reports | (2019)9:7363 | https://doi.org/10.1038/s41598-019-40731-2 1 www.nature.com/scientificreports/ www.nature.com/scientificreports

signifcantly exceeds that found in maternal serum, with slightly less IgG4 being transported. Te trans-placental transport of IgG2 and IgG3 is the least efcient, being roughly equal.21–26. Interestingly, both IgG1 and IgG2 are reported to have similar fractional catabolic rates and half-lives of 21–28 days27,28, indicating that FcRn is able to rescue IgG1 and IgG2 equally efciently from lysosomal degradation and implying that the mechanisms for transplacental transport may difer from those involved in apical recycling. In contrast, the half-life of IgG3 is approximately 7 days (comparable to that of other non-FcRn binding serum proteins), suggesting the FcRn rescuing function to be defcient for IgG327. Despite this, IgG3 is transported across the placenta to a similar extent as IgG2, albeit to a lower degree than IgG121–26. We recently demonstrated that the short half-life and lowered FcRn-meditated transcytosis of IgG3 were due to competition between diferent subclasses for FcRn-mediated rescue and that IgG3 was less successful due to a single amino acid diference within its FcRn-binding site. At this position (435) IgG3 has an arginine compared to histidine in other subclasses28,29. Individuals with the histidine-containing allotype of IgG3 (G3m(s,t) allotype which is uncommon in Europe, but relatively common in Asia and Africa), have half-life and transport across the placenta comparable to IgG125,28,30. Conversely, IgG1 and IgG2 are not known to difer in their afnities for FcRn or known IgG-FcRn contact residues6,31. Other FcγR might be involved in trafcking IgG across the placenta such as FcγRIIb20 and might favour certain subclasses. However, whereas FcRn is required, the role of FcγRIIb has been excluded in the trans-placental transport of mice32,33. Recently, we also found no evidence for any preference for either light chain isotype or IgG2 hinge isomer (formed by diferent disulphide-bridge formations between cysteines in the upper IgG2-hinge and the light chains), indicating that the diference in hinge fexibility reported by Dillon et al.34 does not afect FcRn function26. We have also recently generated a IgG variant lacking Fc-receptor efector functions by engrafing IgG2- and IgG4-derived amino acids onto IgG1 (Δnab35), which still binds FcRn. However, this variant showed surprisingly low FcRn-mediated transport36, warranting further investigation. Here we present a study on FcRn mediated IgG1 and IgG2 transcytosis and describe new fndings on the diferential transport of these subclasses. Our data suggest that IgG2 is transported less efciently by FcRn due to the shorter and less fexible hinge of IgG2 than IgG1 and IgG334,37. We postulate that the shortened hinge may infuence the interaction of IgG2 with FcRn dimers because of steric hindrance with the plasma membrane6,10. Materials and Methods Cell culture. Human choriocarcinoma cells (JAR, American Type Culture Collection, Manassas, VA) were grown in Iscove’s Modifed Dulbecco’s Medium (IMDM) (Cambrex, Verviers, Belgium), and melanoma cells (wild type A375 lacking FcRn expression (but expressing Beta-2-microglobuline (β2m)) and A375 transfected with the FcRn α-chain)28, in Roswell Park Memorial Institute (RPMI) 1640 medium (Invitrogen/Gibco, Carlsbad, California), both supplemented with L-glutamin (300 μg/ml, Invitrogen/Gibco), penicillin (100 U/ml, PAA Laboratories GmbH, Pasching, Germany), streptomycin (100 μg/ml, PAA) and 10% foetal calf serum (FCS, Bodinco, Alkmaar, Te Netherlands). All cultures were carried out at 37 °C, in saturated humidity and 5% CO2 in air.

Isolation of White Blood Cells (WBCs). WBCs were isolated from 9 mL of freshly drawn blood, taken up in VACUETTE heparin blood collection tubes (Greiner Bio-one, Alphen a/d Rijn Nederland). Whole blood was separated by centrifugation at 1600 x g for 10 min, serum fraction was discarded. Red blood cells were lysed by three consecutive steps of hypotonic lysis in cell culture grade water (Gibco) for 30 sec, followed by addition of fltered (0.2 µm, Whatman) 10 x PBS. Te protocol was performed at 4 °C/on ice.

Fluorescence activated cell sorter (FACS). A375 (wild-type and human FcRn) and JAR cells were treated with trypsin Ethylenediaminetetraacetic acid (EDTA) (0.5% and 0.2% (m/V)) for 5 min at culture condi- tions. 2 × 105 cells per well were added to 96 well V-bottom plate (Costar/Corning) and stained with biotinylated mouse anti human Cluster of diferentiation (CD)64 (FcγRI) (BD Pharmingen), CD32 (FcγRII) (Bio-Rad) and CD16 (FcγRIII) (SouthernBiotech, Birmingham, Alabama) for 30 min at 4 °C/on ice. Cells were washed thor- oughly and incubated with Alexa Fluor 633 (AF 633) conjugated Streptavidin (Termo Scientifc). FACS analysis was performed on LSR-II (BD Biosciences), data was analyzed using FlowJo v10 (FlowJo, LLC). Te experiment was performed in duplicate.

IgG. For an overview of the antibodies used in this article and the fgures they were used for, see Table 1. Recombinant B2G antibodies directed against human platelet antigen (HPA)-1a were produced as previously described38. For the B2G mutants, residues in IgG1 were substituted with corresponding amino acids from IgG2 and IgG4, reported to be responsible for their low afnity for FcγR. Te Δc signifes alterations of the IgG1 backbone originating from IgG2 (E233P, L234V, L235A). Δb indicates the same substitution but accompanied by deletion of G236, a residue lacking in IgG2. Δa mutations originate from IgG4 (A327G, A330S, P331S). Afer alteration of three more residues aimed at removing allotypic variation from IgG1 (Δn, K214T, D356E, and L358M), this resulted in two variants of IgG1: Δnab and Δnac difering only by absence or presence of G236, both demonstrating severely reduced binding to FcγR but retaining all residues described to be involved in binding to FcRn35,39,40. Production of the wild type B2G and its variants were performed as described previously. Recombinant V-gene matched IgG1 and IgG2 GDob1 antibodies, directed against Streptococcus pneumoniae serotype 641,42, were produced in 293Freestyle cells (Invitrogen) according to the manufacturer’s instructions. GDob1IgG1H435A was generated as described42. Likewise, GDob1IgG1Δ236 G, GDob1IgG2 + 236 G, and GDob1IgG2H435A, were generated using the Quickchange Site-directed-mutagenesis kit (Stratagene, California, USA) using the following primers and their complementary primers:

Scientific Reports | (2019)9:7363 | https://doi.org/10.1038/s41598-019-40731-2 2 www.nature.com/scientificreports/ www.nature.com/scientificreports

Name Subclass mutations in the backbone specifcity of CDR Used in fgure B2G1 IgG1 none HPA-1 2 K214T, D356E, L358M, A327G, A330S, P331S, B2G1Δnab IgG1 HPA-1 2 E233P, L234V, L235A, deletion of G236 K214T, D356E, L358M, A327G, A330S, P331S, B2G1Δnac IgG1 HPA-1 2 E233P, L234V, L235A B2G2 IgG2 none HPA-1 2 GDob1 IgG1 IgG1 none Streptococcus pneumoniae serotype 6 3, 5, 7 GDob1 IgG1H435A IgG1 H435A Streptococcus pneumoniae serotype 6 3 GDob1 IgG1ΔG236 IgG1 deletion of Gly 236 Streptococcus pneumoniae serotype 6 3, 5, 7 GDob1 IgG2 IgG2 none Streptococcus pneumoniae serotype 6 3, 5, 7 GDob1 IgG2H435A IgG2 H435A Streptococcus pneumoniae serotype 6 3 GDob1 IgG2 + G236 IgG2 insertion of Gly 236 Streptococcus pneumoniae serotype 6 3, 5, 7

Table 1. Nomenclature and mutations of antibodies used in this paper. A legend to the antibodies used in this paper, including the name, subclass, mutations, antigen specifcity and list of fgures in which they were used.

GDob1IgG1Δ236G :5′ CCT CAG CAC CTG AAC TCC TGG G^^ ^AC CGT CAG TCT TCC TCC TCT TC 3′ GDob1IgG2 + 236G :5′ CTC TTC CTC AGC ACC ACC TGT GGC AGG AGG GCC GTC AGT CTT CCT CTT CCC CCC 3′ GDob1IgG2H435A :5′ GAG GCT CTG CAC AAC GCC TAC ACG CAG AAG AGC C 3′. All mutations were confrmed by sequencing (ABI 373 Stretch automated sequencing machine, Applied Biosystems, Foster City, CA) prior to expression.

Surface plasmon resonance (SPR). Human FcRn was produced in-house as described in43 and44, respec- tively. Afnity measurements were performed with SPR using the IBIS MX96 (IBIS Technologies, Enschede, the Netherlands) as described by Dekkers et al.45. Human FcRn was spotted in six duplicates with a three-fold dilution ranging from 30 nM to 1 nM on a SensEye G-streptavadin sensor (Ssens, Enschede, the Netherlands) using a Continuous Flow Microspotter (Wasatch Microfuidics, Salt Lake City, Utah)… Te GDob1 antibodies were injected in a two-fold dilution series starting from 0.49 nM to 125 nM over the FcRn–sensor in PBS-Tween 80 (0.075%) at pH 6.0. Afer every injection a two-step regeneration was performed using PBS pH8.5 + 0.075% Tween-80. Calculation of the dissociation constant (KD) was done using an equilibrium analysis by intrapolation to Rmax = 1000 for FcRn45. Analysis and calculation of all binding data were carried out with Scrubber sofware version 2 (Biologic Sofware, Campbell, ACT, Australia) and Microsof Ofce Excel 2013.

IgG transcytosis. Transcytosis experiments with A375 (wild-type and human FcRn) and JAR cells were performed as previously described28. Briefy, 12 mm polycarbonate Transwell flters (0.4 μm pore size, Costar/ Corning) were inoculated with 5 × 105 cells, grown overnight to confuence, washed with phosphate bufered saline (PBS) and medium replaced with fresh medium (IMDM at pH 7.4 with supplements as stated above). Mixtures of IgG contained streptavidin-horseradish peroxidase (HRP) (Sanquin) to assess background transport. Apical to basolateral transport was calculated according to ([IgG]basolateral × 1.5 ml)/([IgG] input × 0.5 ml) × 100%. All experiments were performed in triplicate.

Serum samples. Serum samples from the umbilical cord of three newborns and matched serum samples from mothers were drawn at birth. IgG subclass levels in these samples were determined by nephelometry as described below. Signed informed consent was obtained from all women, and the collection of blood samples and clinical data received approval by the Ethics Committee of the Leiden University Medical Centre (P02–200) as has been previously reported24, in accordance with relevant guidelines.

IgG quantifcation. IgG subclass concentrations in serum samples were determined by Nephelometry (Behringer Nephelometer II, Behringer diagnostics, Deerfeld, Illinois, USA) according to manufacturer’s pro- tocols. In all in vitro studies, IgG subclasses were quantifed by sandwich enzyme-linked immunosorbent assay (ELISA) using subclass specifc mouse monoclonal antibodies (IgG1: MH161-1; IgG2: HP6062, Sanquin) to cap- ture and a directly HRP conjugated monoclonal mouse anti-IgG (JDC-10, Southern Biotech, Birmingham, AL, USA) for detection. Conversion of 3,3′,5,5′-Tetramethylbenzidine (TMB) was used to quantitate HRP activity per well and absorptions were read using a Sunrise TECAN spectrophotometer. Te concentrations were read from a standard curve made from the IgG preparations used for transport (in vitro studies) or IVIg.

High-Pressure Liquid Chromatography Size-exclusion chromatography (HPLC-SEC). ÄKTA explorer P900 (GE Healthcare) equipped with Superdex 200 10/300GL (GE Healthcare) was employed for HPLC-SEC. Te column was adjusted to previously degassed PBS with two column volumes prior to sample run. 100 µL of purifed Ab at a concentration of 1 mg/mL was injected manually. Analysis was performed at a fow of 0.5 mL/min, detecting at UV 280 nm.

Scientific Reports | (2019)9:7363 | https://doi.org/10.1038/s41598-019-40731-2 3 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 1. Relative IgG subclass concentrations in cord blood and maternal serum. Te relative subclass concentration in cord blood as compared to the concentrations found in maternal serum, expressed as average cord/mother ratios for each IgG subclass. Data are from three paired mother – child samples and expressed as means plus standard deviation. All data are from 3 independent experiments, expressed as mean plus standard deviation. Data was analysed by one-way ANOVA with Tukey’s multiple comparison test and signifcance is shown as previously indicated.

Statistical analysis and data sets. All data represent the mean and standard deviation of at least three independent experiments. All transcytosis assays consisted of three replicates. GraphPad Prism for Windows (GraphPad Sofware) was used for all statistical analysis. Signifcance was set at P < 0.05, and is indicated on all fgures as *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.

“Informed Consent”. Signed informed consent was obtained from all women who donated serum sample of themselves and cord blood of their new-borns. Te collection of blood samples and clinical data received approval by the Ethics Committee of the Leiden University Medical Centre (P02–200). Results IgG2 and IgG3 are transported across the placenta to a lesser extent than IgG1 and IgG4. We analyzed three matched mother and cord sera at term. Similar to previous studies21,23–26, we found that only IgG1 had a higher concentration in cord blood than in maternal serum, IgG2 and IgG3 were present in lower amounts in cord blood than in maternal serum, and IgG4 concentrations were equal (Fig. 1). FcRn mediated apical to basolateral IgG1 transport is more efcient than IgG2 transport in vitro. In order to analyze the relative transport of IgG1 and IgG2 and the role of FcγR and FcRn, we analyzed their transport in vitro using the choriocarcinomic JAR cells (endogenously expressing FcRn)28, A375 cells devoid of FcRn28 expression, and A375- FcRn transfectants46. None of these cell lines expressed classical FcγR as seen by FACS (Supplementary Fig. 1). As seen in vivo, we observed that IgG1 (B2G1) was transported more efciently than IgG2 (B2G2) (an otherwise identical antibody of the IgG2 subclass) both by JAR cells (Fig. 2A) and A375-FcRn cells (Fig. 2B), but not FcRn-defcient wild type A375 cells (Fig. 2C).

IgG transcytosis is independent of the ability to bind Fcγ receptors. We have previously described B2G1Δnac, a mutated IgG1 variant with amino acids which are implicated in FcγR binding being replaced by corresponding residues found in IgG2 or IgG4. Tis variant displays almost no binding to activating FcγR but retains 40% binding to the inhibitory FcγRIIb39,47. Tis mutant was transported with equal efciency as wild type IgG1 antibody in JAR cells (Fig. 2A) and A375-FcRn cells (Fig. 2B). However, another variant, B2G1Δnab, identical and with similar FcγR-binding properties as B2G1Δnac but additionally lacking glycine 236 (G236) which is absent in wild type IgG2, was transported less efciently and to the same extent as the wild type IgG2 molecule (B2G2) (Fig. 2A,B). Tis was observed in both JAR (Fig. 2A) and A375-FcRn cells (Fig. 2B) but no signifcant transport was found in wild type (WT)A375 cells lacking FcRn and FcγRs expression28 (Fig. 2C). Tese fndings demonstrated that FcRn, and not classical FcγR, was required for transcytosis of IgG in both JAR and A375 cells and suggested that absence of G236 may reduce the efciency of this transcytosis.

Apical to basolateral transcytosis of IgG is influenced by glycine 236. To ascertain whether the reduced IgG2 transport was due to the lack of G236 compared to IgG1, we generated recombinant antibodies difering only in this single amino acid. Afnity measurements by SPR with c-terminally biotinylated FcRn on streptavidin sensors showed that either removing G236 from IgG1 or introducing it into IgG2, did not apparently infuence binding to FcRn on the chip (Fig. 3). Te wild-type and mutant GDob1 antibodies displayed a similar minor fraction of dimers which was similar across all variants (Supplementary Fig. 2), also seen in rather steep associations and slow disassociation in Fig. 3. We found that IgG1 lacking only G236 (GDob1IgG1ΔG236) was transported to an equal level as wild type IgG2 (Fig. 4A). Moreover, IgG2 with G236 inserted (GDob1IgG2 + G236) demonstrated enhanced transport and was transported to an equal level as wild

Scientific Reports | (2019)9:7363 | https://doi.org/10.1038/s41598-019-40731-2 4 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 2. FcRn-mediated IgG transport is independent of FcγR and is infuenced by the presence or absence of G236. Transport of IgG1 variants B2G1, B2G1Δnac and B2G1Δnab as well as IgG2 B2G2 was tested using diferent cell lines. (A) placenta derived cells (JAR). (B) A375-FcRn cells, transfected with the FcRn alpha chain. (C) 375 wild type cells lacking functional FcRn expression. All experiments were run for two hours and transport was from the apical to the basolateral side of a monolayer of cells. HRP was included as a measure of aspecifc leakage and measured in the same samples as IgG. All data are from 3 independent experiments, expressed as mean plus standard deviation. Data was analysed by one-way ANOVA with Tukey’s multiple comparison test and signifcance is shown as previously indicated

type IgG1, confrming that the lack of G236 in IgG resulted in its less efcient FcRn-mediated transcytosis. When the afnity of the primary binding site of IgG for FcRn was strongly reduced by changing the histidine in position 435 to an alanine (H435A)48,49 signifcant transport was still observed, but it was severely reduced as expected. Importantly, no signifcant diference between IgG1 and IgG2 transport was observed when both carried the H435A mutation, suggesting FcRn-mediated efector functions were responsible for the observed diferences

Scientific Reports | (2019)9:7363 | https://doi.org/10.1038/s41598-019-40731-2 5 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 3. Mutant GDob1IgG1ΔG236 and GDob1IgG2 + G236 retain afnities to human FcRn in comparison to WT antibodies. Sensorgrams obtained from afnity measurement of GDob1IgG1, GDob1IgG1ΔG236, GDob1IgG2 and GDob1IgG2 + G236 to human FcRn in SPR. Antibodies were injected at concentrations ranging from 125 nM to 0.49 nM in two-fold dilutions over biotinylated human FcRn coupled to a streptavidin biosensor at pH 6.0. Te afnities (M) calculated from afnity plots derived from SPR measurements of GDob1 IgG1, GDob1 IgG1 ΔG236, GDob1 IgG2 and GDob1 IgG2 + G236 to human FcRn. KDs were calculated using equilibrium analysis by intrapolation to Rmax = 1000 human FcRn. (−/−).

in IgG1 and IgG2 transport efciency. In addition, role of FcγR can be excluded by their absence in these cells (Supplemental Fig. 1).

Reduced FcRn mediated transport correlates with enhanced degradation. To investigate whether the reduced transcytosis rates of IgG without G236 were due to increased degradation, we sampled both apical and basolateral compartments afer 18 hours of transport. For both wild type GDob1IgG1 and GDob1IgG2 + G236 the total amounts detected were in excess of 95% of the starting material, while wild type GDob1IgG2 and GDob1IgG1ΔG236 were recoverable to a signifcantly lesser extent (Fig. 4B). Tis suggested that the reduced FcRn-mediated transcytosis of GDob1IgG2 and GDob1IgG1ΔG236 (Figs 1–2) was due to enhanced degradation. Discussion Te long half-life of IgG and its transplacental transport are both mediated by the neonatal Fc receptor5,19,23. FcRn-mediated half-life of IgG2 is comparable to that of IgG127 but transport across the human placenta is lower for IgG221–26, suggesting diferences between the mechanisms of FcRn-mediated recycling and transcytosis of IgG. Te reason for this is unclear, as IgG1 and IgG2 have not been reported to difer in residues found to be important for IgG-FcRn interaction6 and their binding to FcRn immobilized on a biosensor chip were not reported to be signifcantly diferent31. However, the stoichiometry10, the lateral fuidity of transmembrane receptors, and the physiological proximity of the plasma membrane cannot be accounted for on a surface like plasmon resonance biosensor chips, and therefore, such assays do not always accurately represent the biological context. Te lowered in vitro FcRn-mediated transcytosis rates of IgG2 compared to IgG1 which we observed were in line with the reported diferences in maternal-cord blood ratios. Tis diference was solely attributed to FcRn-mediated transcytosis as transcytosis of B2G1Δnac, an IgG1 variant unable to bind to classical FcγR38,39,47, was unafected in both JAR and A375-FcRn cells. Unexpectedly, the transcytosis rate of another IgG1 variant, B2G1Δnab, was reduced to the level of IgG2 transcytosis. Te only diference between the Δnac and Δnab variants was the deletion of G236 in the latter, similar to IgG2 that also lacks G236. Using unrelated human IgG1 and IgG2 VH-matched GDob1 anti- bodies41,42 with identical afnity for FcRn in SPR experiments, we confrmed that G236, but not any other mutations in B2G1Δnac or B2G1Δnab, infuenced the transport of IgG. Interestingly, G236 has not previously been reported to afect FcRn function and is located in the lower hinge, far from the FcRn binding site6 (Fig. 5). Gurbuxani et al. postulated a mathematical model for FcRn-IgG interaction50,51, in which either FcRn can bind to IgG in two distinct modes, or IgG might have two FcRn binding sites, functioning synergistically. Whether such a secondary interaction-domain within IgG exists is unclear, but it cannot be excluded as a cocrystal of a complete IgG-FcRn complex has not yet been resolved. Björkman and colleagues observed an FcRn dimer both in crystals of rat FcRn and in the cocrystal of FcRn with Fc6,49. Later work demonstrated that Fc and FcRn can crystallize in oligomeric ribbons with a symmetrical repeating unit of 2 FcRn:1 IgG (FcRn:IgG:FcRn) with the C-terminus of IgG orientated towards the N-terminus of FcRn52. Both the proposed “lying down” or “standing up” models for FcRn engaged with IgG binding have difculties explaining how a complete IgG binds to FcRn in a cellular context. In the “standing up” model the two Fab fragments would be projected to either protrude into the plasma membrane, or, due to the fexibility of the hinge, bend back (Fig. 5A). Te fragment antigen binding (Fab) might be more easily accommodated if FcRn is tilted towards the cell but some bending at the hinge would still be required (Fig. 5B). Te binding of a second FcRn to this complex, forming a 2:1 complex of FcRn and IgG which might be important for FcRn-signalling events16, may occur within the IgG/FcRn positive-transport tubules observed by the groups of Ober and Ward et al.53,54. Within these tubules one molecule of IgG may be bound by two FcRn molecules residing in a tilted orientation on parallel membranes. Tis is similar to what was suggested to occur between adjacent microvilli in the rodent gut by Bjorkman et al.6, who observed by electron cryotomography that IgG-Fc complexes assemble preferentially on interfaces between layers of FcRn expressing membranes54,55. Whether this “laying down” model, the “standing up” model or both is more relevant for physical transport of IgG is currently unknown.

Scientific Reports | (2019)9:7363 | https://doi.org/10.1038/s41598-019-40731-2 6 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 4. Low transport of IgG2 was due to lack of G236 in IgG2, and resulted in enhanced degradation. (A) Apical to basolateral transport of VH-matched GDob1 wild type and ΔG236 IgG1 variants as well as IgG2 wild type, and + G236 variants by A375 FcRn cells. GDob1IgG1H435A and GDob1IgG2H435A were included as control in which the main FcRn binding pocket was disrupted and HRP was included as a measure of passive difusion. Experiments were run for 2 hours. (B) Afer 18 hours of apical to basolateral transport in A375-FcRn cells, both apical (hatched bars) and basolateral (open bars) compartments were sampled and IgG concentrations were determined. Approximately 95% of GDob1IgG1 and GDob1IgG2 + G236 could be accounted for, while only about 80% of GDob1IgG1Δ236Gly and GDob1IgG2 was detectable. Data shown are from three independent experiments, expressed as mean plus standard deviations. In B) the total values from apical and basolateral samples taken from the same transwells. Signifcance was tested by one way ANOVA with Tukeys multiple comparisons test in both (A,B). In (A) statistical comparison within an IgG subclass is shown without brackets and only shown for comparison with the WT variant, but between subclasses (with brackets) only shown for WT subclasses, G236 inserted in IgG2 and G236 removed from IgG1, and matched for the presence or absence of G236 IgG1 and IgG2 variants.

G236 is present in the lower hinge of all subclasses except IgG2, and increases the fexibility and extends the distance of the two Fab domains from the Fc portion. Based on the structural constraints for binding of whole IgG by FcRn, and the fndings presented here, we predict that the shortened hinge of IgG2 makes the formation of an FcRn:IgG2:FcRn complex less energetically favourable than is the case for other IgG and thus decreases the compatibility of IgG2 with FcRn binding and transport. Tis is supported by recent evidence showing that the amino-acid composition of the Fab fragment and charge in particular efects FcRn binding and recycling56. Our study may in part provide a solution as to why IgG2 transcytosis takes place at a lower rate than IgG1.

Scientific Reports | (2019)9:7363 | https://doi.org/10.1038/s41598-019-40731-2 7 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 5. IgG binds FcRn in a top down orientation. IgG1 is depicted in red and blue colours (the two light chains in red and light red, and the two heavy chains in blue and light blue). Te α-chain of FcRn is depicted in orange, and β2M in yellow and is modelled on a plasma membrane. Te positions of the two potential FcRn binding sites on the IgG1 are indicated by showing the critical histidine residues in green and their position is indicated with arrows in (A) (upper arrow H435 in CH3; lower arrow H310 in CH2). One of the binding sites is unoccupied. Te position of G236 is indicated in yellow and with arrows in (B). In (A), a model of FcRn binding to human IgG1 shows that the Fab portions of the molecule will clash into the plasma membrane if FcRn is extending directly from in the plasma membrane. Tis could be avoided if the Fab fragments bend back at the hinge and/or FcRn bends back against the plasma membrane, as in (B). Data are based on the crystal structure of rat FcRn with Fc (Accession number 1I1A Ref martin et al. mol cell 2001) and overlaid with the structure of human IgG1 (Accession number 1HZH, Saphire 2001 Science). UCSF Chimera (Pettersen, et al. 2004) was used for modelling and imaging.

However, it does not explain why in vivo rates of IgG2 and IgG1 catabolism do not seem to difer, unless these two FcRn-mediated processes fundamentally difer, or hitherto unknown factors make IgG2 less prone to degradation, compensating for the reduced FcRn rescue function. Tis last option is however entirely speculative and we are not aware of any data indicating that this is the case. In summary, we show that the lowered transplacental transport of IgG2 in vivo can be simulated in an in vitro transcytosis system, is FcRn- but not FcγR-mediated, and is caused by the absence of G236 in IgG2. More efort is needed to unravel the cellular mechanism behind the observation that the short lower hinge of IgG2 afects transcytosis but not its half-life.

Scientific Reports | (2019)9:7363 | https://doi.org/10.1038/s41598-019-40731-2 8 www.nature.com/scientificreports/ www.nature.com/scientificreports

References 1. Jones, E. A. & Waldmann, T. A. Te mechanism of intestinal uptake and transcellular transport of IgG in the neonatal rat. J Clin Invest 51, 2916–2927 (1972). 2. Simister, N. E. & Mostov, K. E. An Fc receptor structurally related to MHC class I antigens. Nature 337, 184–187 (1989). 3. Simister, N. E. & Rees, A. R. Isolation and characterization of an Fc receptor from neonatal rat small intestine. Eur J Immunol 15, 733–738 (1985). 4. Simister, N. E., Story, C. M., Chen, H. L. & Hunt, J. S. An IgG-transporting Fc receptor expressed in the syncytiotrophoblast of human placenta. Eur J Immunol 26, 1527–1531 (1996). 5. Story, C. M., Mikulska, J. E. & Simister, N. E. A major histocompatibility complex class I-like Fc receptor cloned from human placenta: possible role in transfer of immunoglobulin G from mother to fetus. J Exp Med 180, 2377–2381 (1994). 6. Burmeister, W. P., Huber, A. H. & Bjorkman, P. J. Crystal structure of the complex of rat neonatal Fc receptor with Fc. Nature 372, 379–383 (1994). 7. Popov, S. et al. Te stoichiometry and afnity of the interaction of murine Fc fragments with the MHC class I-related receptor, FcRn. Mol Immunol 33, 521–530 (1996). 8. Raghavan, M., Gastinel, L. N. & Bjorkman, P. J. The class I major histocompatibility complex related Fc receptor shows pH- dependent stability diferences correlating with immunoglobulin binding and release. Biochemistry 32, 8654–8660 (1993). 9. Vaughn, D. E. & Bjorkman, P. J. Structural basis of pH-dependent antibody binding by the neonatal Fc receptor. Structure (London, England: 1993) 6, 63–73 (1998). 10. Raghavan, M., Wang, Y. & Bjorkman, P. J. Effects of receptor dimerization on the interaction between the class I major histocompatibility complex-related Fc receptor and IgG. Proc Natl Acad Sci USA 92, 11200–11204 (1995). 11. Wu, Z. & Simister, N. E. Tryptophan- and dileucine-based endocytosis signals in the neonatal Fc receptor. J Biol Chem 276, 5240–5247 (2001). 12. McCarthy, K. M. et al. Efects of mutations in potential phosphorylation sites on transcytosis of FcRn. J Cell Sci 114, 1591–1598 (2001). 13. McCarthy, K. M., Yoong, Y. & Simister, N. E. Bidirectional transcytosis of IgG by the rat neonatal Fc receptor expressed in a rat kidney cell line: a system to study protein transport across epithelia. J Cell Sci 113, 1277–1285 (2000). 14. Wernick, N. L., Haucke, V. & Simister, N. E. Recognition of the tryptophan-based endocytosis signal in the neonatal Fc Receptor by the mu subunit of adaptor protein-2. J Biol Chem 280, 7309–7316 (2005). 15. Newton, E. E., Wu, Z. & Simister, N. E. Characterization of basolateral-targeting signals in the neonatal Fc receptor. J Cell Sci 118, 2461–2469 (2005). 16. Tesar, D. B., Tiangco, N. E. & Bjorkman, P. J. Ligand valency afects transcytosis, recycling and intracellular trafcking mediated by the neonatal Fc receptor. Trafc 7, 1127–1142 (2006). 17. Lencer, W. I. & Blumberg, R. S. A passionate kiss, then run: exocytosis and recycling of IgG by FcRn. Trends Cell Biol 15, 5–9 (2005). 18. Junghans, R. P. & Anderson, C. L. Te protection receptor for IgG catabolism is the beta2- microglobulin-containing neonatal intestinal transport receptor. Proc Natl Acad Sci USA 93, 5512–5516 (1996). 19. Roopenian, D. C. & Akilesh, S. FcRn: the neonatal Fc receptor comes of age. Nat Rev Immunol (2007). 20. Lyden, T. W. et al. Te Fc receptor for IgG expressed in the villus endothelium of human placenta is Fc gamma RIIb2. J Immunol 166, 3882–3889 (2001). 21. Malek, A., Sager, R., Kuhn, P., Nicolaides, K. H. & Schneider, H. Evolution of maternofetal transport of immunoglobulins during human pregnancy. Am J Reprod Immunol 36, 248–255 (1996). 22. Garty, B. Z., Ludomirsky, A., Danon, Y. L., Peter, J. B. & Douglas, S. D. Placental transfer of immunoglobulin G subclasses. Clin Diagn Lab Immunol 1, 667–669 (1994). 23. Simister, N. E. Placental transport of immunoglobulin G. Vaccine 21, 3365–3369 (2003). 24. Einarsdottir, H. K. et al. Comparison of the Fc glycosylation of fetal and maternal immunoglobulin G. Glycoconjugate journal 30, 147–157, https://doi.org/10.1007/s10719-012-9381-6 (2012). 25. Einarsdottir, H. et al. H435-containing immunoglobulin G3 allotypes are transported efficiently across the human placenta: implications for alloantibody-mediated diseases of the newborn. Transfusion 54, 665–671, https://doi.org/10.1111/trf.12334 (2014). 26. Einarsdottir, H. K. et al. On the Perplexingly Low Rate of Transport of IgG2 across the Human Placenta. PLoS One 9, e108319, https://doi.org/10.1371/journal.pone.0108319 (2014). 27. Morell, A., Terry, W. D. & Waldmann, T. A. Metabolic properties of IgG subclasses in man. J Clin Invest 49, 673–680 (1970). 28. Stapleton, N. M. et al. Competition for FcRn-mediated transport gives rise to short half-life of human IgG3 and ofers therapeutic potential. Nat Commun 2, 599, https://doi.org/10.1038/ncomms1608 (2011). 29. Kim, J. K. et al. Mapping the site on human IgG for binding of the MHC class I-related receptor, FcRn. Eur J Immunol 29, 2819–2825 (1999). 30. Dechavanne, C. et al. Associations between an IgG3 polymorphism in the binding domain for FcRn, transplacental transfer of malaria-specifc IgG3, and protection against Plasmodium falciparum malaria during infancy: A birth cohort study in Benin. PLoS Med 14, e1002403, https://doi.org/10.1371/journal.pmed.1002403 (2017). 31. West, A. P. Jr. & Bjorkman, P. J. Crystal structure and immunoglobulin G binding properties of the human major histocompatibility complex-related Fc receptor. Biochemistry 39, 9698–9708 (2000). 32. Mohanty, S. et al. IgG is transported across the mouse yolk sac independently of FcγRIIb. Journal of reproductive immunology 84, 133–144, https://doi.org/10.1016/j.jri.2009.10.008 (2010). 33. Kim, J. et al. FcRn in the yolk sac endoderm of mouse is required for IgG transport to fetus. Journal of immunology (Baltimore, Md.: 1950) 182, 2583–2589, https://doi.org/10.4049/jimmunol.0803247 (2009). 34. Dillon, T. M. et al. Structural and Functional Characterization of Disulfde Isoforms of the Human IgG2 Subclass. J Biol Chem 283, 16206–16215 (2008). 35. Armour, K. L., Clark, M. R., Hadley, A. G. & Williamson, L. M. Recombinant human IgG molecules lacking Fcgamma receptor I binding and monocyte triggering activities. Eur J Immunol 29, 2613–2624 (1999). 36. Stapleton, N. M. et al. Human IgG lacking efector functions demonstrate lower FcRn-binding and reduced transplacental transport. Mol Immunol 95, 1–9, https://doi.org/10.1016/j.molimm.2018.01.006 (2018). 37. Wypych, J. et al. Human IgG2 Antibodies Display Disulfde-mediated Structural Isoforms. J Biol Chem 283, 16194–16205 (2008). 38. Joutsi-Korhonen, L. et al. Te efect of recombinant IgG antibodies against the leucine-33 form of the platelet beta3 integrin (HPA- 1a) on platelet function. Trombosis and haemostasis 91, 743–754 (2004). 39. Armour, K. L., van de Winkel, J. G., Williamson, L. M. & Clark, M. R. Differential binding to human FcgammaRIIa and FcgammaRIIb receptors by human IgG wildtype and mutant antibodies. Mol Immunol 40, 585–593 (2003). 40. Ghevaert, C. et al. Developing recombinant HPA-1a-specifc antibodies with abrogated Fcgamma receptor binding for the treatment of fetomaternal alloimmune thrombocytopenia. J Clin Invest 118, 2929–2938, https://doi.org/10.1172/jci34708 (2008). 41. Saeland, E. et al. Central Role of Complement in Passive Protection by Human IgG1 and IgG2 Anti-pneumococcal Antibodies in Mice. J Immunol 170, 6158–6164 (2003). 42. Vidarsson, G. et al. FcRn: an IgG receptor on phagocytes with a novel role in phagocytosis. Blood 108, 3573–3579 (2006). 43. Dekkers, G. et al. Decoding the Human Immunoglobulin G-Glycan Repertoire Reveals a Spectrum of Fc-Receptor- and Complement-Mediated-Efector Activities. Frontiers in immunology 8, 877, https://doi.org/10.3389/fmmu.2017.00877 (2017).

Scientific Reports | (2019)9:7363 | https://doi.org/10.1038/s41598-019-40731-2 9 www.nature.com/scientificreports/ www.nature.com/scientificreports

44. Schlothauer, T. et al. Analytical FcRn afnity chromatography for functional characterization of monoclonal antibodies. mAbs 5, 576–586, https://doi.org/10.4161/mabs.24981 (2013). 45. Dekkers, G. et al. Afnity of human IgG subclasses to mouse Fc gamma receptors. MAbs 9, 767–773, https://doi.org/10.1080/19420 862.2017.1323159 (2017). 46. Cassard, L. et al. Expression of low-afnity Fc gamma receptor by a human metastatic melanoma line. Immunol Lett 75, 1–8 (2000). 47. Armour, K. L. et al. Intravascular survival of red cells coated with a mutated human anti-D antibody engineered to lack destructive activity. Blood 107, 2619–2626, https://doi.org/10.1182/blood-2005-03-0989 (2006). 48. Shields, R. L. et al. High resolution mapping of the binding site on human IgG1 for Fc gamma RI, Fc gamma RII, Fc gamma RIII, and FcRn and design of IgG1 variants with improved binding to the Fc gamma R. J Biol Chem 276, 6591–6604 (2001). 49. Burmeister, W. P., Gastinel, L. N., Simister, N. E., Blum, M. L. & Bjorkman, P. J. Crystal structure at 2.2 A resolution of the MHC- related neonatal Fc receptor. Nature 372, 336–343 (1994). 50. Gurbaxani, B., Dela Cruz, L. L., Chintalacharuvu, K. & Morrison, S. L. Analysis of a family of antibodies with diferent half-lives in mice fails to fnd a correlation between afnity for FcRn and serum half-life. Mol Immunol 43, 1462–1473 (2006). 51. Gurbaxani, B. M. & Morrison, S. L. Development of new models for the analysis of Fc-FcRn interactions. Mol Immunol 43, 1379–1389 (2006). 52. Martin, W. L., West, A. P. Jr., Gan, L. & Bjorkman, P. J. Crystal structure at 2.8 A of an FcRn/heterodimeric Fc complex: mechanism of pH-dependent binding. Mol Cell 7, 867–877 (2001). 53. Ober, R. J., Martinez, C., Vaccaro, C., Zhou, J. & Ward, E. S. Visualizing the site and dynamics of IgG salvage by the MHC class I-related receptor, FcRn. J Immunol 172, 2021–2029 (2004). 54. He, W. et al. FcRn-mediated antibody transport across epithelial cells revealed by electron tomography. Nature 455, 542–546 (2008). 55. He, Y., Jensen, G. J. & Bjorkman, P. J. Nanogold as a specifc marker for electron cryotomography. Microscopy and microanalysis: the ofcial journal of Microscopy Society of America, Microbeam Analysis. Society, Microscopical Society of Canada 15, 183–188, https:// doi.org/10.1017/s1431927609090424 (2009). 56. Piche-Nicholas, N. M. et al. Changes in complementarity-determining regions signifcantly alter IgG binding to the neonatal Fc receptor (FcRn) and pharmacokinetics. MAbs 10, 81–94, https://doi.org/10.1080/19420862.2017.1389355 (2018). Acknowledgements Ir. N.M.S was supported by an Netherlands Organisation for Scientifc Research (NWO) Aspasia grant number 015.001.083, A.R.T. was supported by an LSBR grant, number 1527 awarded to G.V., S.W.T. was fnancially supported by Genmab, and M.B. by argenx. Molecular graphics images were produced using the UCSF Chimera package from the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco (supported by NIH P41 RR-01081). Author Contributions N.M.S. performed all transwell experiments and ELISA on sera. M.B. and A.E.H.B. performed SPR, FACS and HPLC experiments. A.R.T. and S.W.d.T. generated recombinant FcRn, and W.J.E.E., J.M., GB. and M.M. Biotinylated this product. N.M.S., K.L.A., M.R.C., L.M.W., C.E.vd.S. and G.V. conceived the project, developed the hypothesis and designed the experiments. Figures 1, 2 and 4 were prepared by N.M.S. with assistance from G.V., Molecular modelling and Fig. 5 was prepared by G.V. Figure 3 and Supplemental Figs 1 and 2 were prepared by M.B. and A.E.H.B. Te manuscript was written by N.M.S. and G.V., with assistance from M.B., A.E.H.B. and K.L.A. All authors reviewed the manuscript. Additional Information Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-019-40731-2. Competing Interests: Te authors declare no competing interests. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. Te images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Scientific Reports | (2019)9:7363 | https://doi.org/10.1038/s41598-019-40731-2 10